US6065085A

Movatterモバイル変換

Info

Publication number: US6065085A
Application number: US09/013,818
Authority: US
Inventors: Louis H. Odenwald, Jr.; Steven R. Schremmer
Original assignee: LSI Logic Corp
Current assignee: Avago Technologies International Sales Pte Ltd
Priority date: 1998-01-27
Filing date: 1998-01-27
Publication date: 2000-05-16
Anticipated expiration: 2018-01-27
Also published as: WO1999038086A3; WO1999038086A2; US6223240B1

Abstract

The method and apparatus provides a data processing system. The data processing system includes a primary bus, a secondary bus, and a host processor connected to the primary bus. The data processing system includes a first secondary processor connected to the primary bus and the secondary bus. Additionally, a second secondary processor is connected to the secondary bus. The first secondary processor and the second secondary processor forms cascaded processors for input/output functions. Selected functions normally performed by the second secondary processor are performed by the first secondary processor, wherein a division of workload increases performance of the data processing system. This architecture allows shifting of workload down to the secondary bus.

Description

BACKGROUND OF THE INVENTION

1. Technical Field

The present invention relates generally to an improved data processing system and in particular to a method and apparatus for an improved bus architecture in a data processing system. Still more particularly, the present invention relates to an improved method and apparatus for transferring data between devices on different buses within a data processing system.

2. Description of the Related Art

The use of a microprocessor and its associated bus architecture in a computer system is well known. Also well known is the coupling of peripheral components onto a bus for providing various other functions related to the computer system. Some examples of such devices are disk drives, disk drive controllers, graphics accelerators, audio cards, modems and network connections. Generally, these peripheral components are coupled to a system component bus, also referred to as a "secondary bus", for effecting data transfer between various components. Typically, with simpler computer systems, the data transfer is between the peripheral device and either the processor (CPU) or main memory.

In high performance computer systems, peripheral component interconnect (PCI) buses are often employed. The PCI bus is a high performance expansion bus architecture that was designed to replace the traditional ISA bus (Industry Standard Architecture bus) and EISA bus (Extended Industry Standard Architecture bus) buses found in many X86 based personal computers. A group of companies (including Intel, IBM, Compaq, DEC, Dell, NEC, etc.) cooperated in preparing and publishing a standard specification for the PCI bus. The specifications for the PCI bus is available from the PCI Special Interest Group, 5200 Elam Young Parkway, Hillsboro, Oreg.

In expanding the capacity of standard PCI, a "bus bridge", also referred to as a "PCI-to-PCI bridge", is employed. The primary function of such a bus bridge is to allow transactions to occur between a master device on one PCI bus and a target device on the other PCI bus. In this manner, the bus bridge provides system and adapter card designers an ability to overcome an electrical loading limits inherent in a standard PCI bus. A bus bridge has two interfaces, a primary interface and a secondary interface. The primary interface is the PCI interface of the bus bridge that is connected to the PCI bus closest to the central processing unit (CPU). The secondary interface is the PCI interface of the bus bridge that is connected to the PCI bus that is farthest from the CPU. Under the PCI bridge specifications, each of the interfaces are capable of either a master or target operation. With respect to the bus which initiates an operation, the bus bridge functions as a target on behalf of the target device that actually resides on the target bus. Likewise, with respect to the target bus, the bus bridge functions as a master on behalf of the master device that actually resides on the initiating bus.

In FIG. 1, a block diagram of a known data processing system architecture is illustrated.Data processing system 100 is a data processing system following the Intelligent Input/Output (I₂ O) Architecture Specification, version 1.5, March 1997 available from the I₂ O Special Interest Group. The specification describes an open architecture for developing device drivers in network system environments. The architecture is independent of the operating system (OS), processor platform, and system I/O bus.

Data processing system 100 is an example of an intelligent input/output (I₂ O) architecture which includes a host central processing unit (CPU) 102 which could be a Pentium processor that is available from Intel Corporation located in Santa Clara, Calif. In an I₂ O architecture, thehost CPU 102 is responsible for running the various operating system modules (OSMs). In this example, theprimary bus 104 is a PCI bus and provides communication between thehost CPU 102, an input/output processor (IOP)/PCI bridge 106 andIOP 108. The IOP/PCI bridge 106 is intended to be a I₂ O core compliant device such as Intel's i960. The IOP/PCI bridge 106 also contains a PCI-to-PCI bridge which bridges theprimary bus 104 to thesecondary bus 110, which in this example is also a PCI bus.Adapter #1 112 andadapter #2 114 are also attached to thesecondary bus 110.

Adapters

112 and 114 are non-intelligent, meaning that they contain no processor, and are reliant on external processing power to run their device drivers, or hardware driver module (HDM) in I₂ O terminology. In this example, both of these adapters' (112 and 114) HDMs are actually running on the IOP/PCI bridge 106. IOP 108 is an example of an I₂ O shell compliant device and is an intelligent adapter such as Symbios' SYMFC920. IOP 108 is specifically designed as an integral part ofadapter #3 and is responsible for runningadapter #3's HDM.

In a data processing system, such as the example system described in FIG. 1, in which there resides an IOP/PCI bridge and one or more additional IOPs it would be prudent to architect the system in such a way as to utilize more than one IOP on a given I/O transaction. Although bus bridges solve electrical loading limits inherent in standard PCI buses and I₂ O offloads many I/O related tasks from the host CPU down to the various IOPs, there are still architectural issues when dealing with multiple inter-operating IOPs. Therefore, it would be advantageous to develop a simple method for allowing IOPs to inter-operate, share resources and distribute the workload in such systems.

SUMMARY OF THE INVENTION

The present invention provides a data processing system. The data processing system includes a primary bus, a secondary bus, and a host processor connected to the primary bus. The data processing system includes a first secondary processor/bridge connected to the primary bus and the secondary bus. Additionally, a second secondary processor is connected to the secondary bus. The first secondary processor/bridge and the second secondary processor forms cascaded processors for input/output functions. Selected functions normally performed by the second secondary processor are performed by the first secondary processor/bridge, wherein a division of workload increases performance of the data processing system.

The first secondary processor/bridge also provides communication between the primary bus and the secondary bus. Through the secondary processor, workload may be shifted to the secondary bus increasing the bandwidth on the primary bus.

BRIEF DESCRIPTION OF THE DRAWINGS

The novel features believed characteristic of the invention are set forth in the appended claims. The invention itself however, as well as a preferred mode of use, further objects and advantages thereof, will best be understood by reference to the following detailed description of an illustrative embodiment when read in conjunction with the accompanying drawings, wherein:

FIG. 1 is a block diagram of a known data processing system architecture;

FIG. 2 is a block diagram of a data processing system in accordance with a preferred embodiment of the present invention;

FIG. 3 is a block diagram of software units within portions of the data processing system in accordance with a preferred embodiment of the present invention;

FIGS. 4A and 4B are flowcharts of a process for loading and initializing a bridge hardware device module in accordance with a preferred embodiment of the present invention;

FIGS. 5A and 5B are flowcharts of a process for messaging handling in accordance with a preferred embodiment of the present invention;

FIG. 6 is a block diagram of a RAID implementation in a data processing system in accordance with a preferred embodiment of the present invention; and

FIG. 7 is a block diagram of a data processing system in accordance with a preferred embodiment of the present invention.

FIG. 8 is a block diagram of a bus bridge in which a preferred embodiment of the present invention may be implemented.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

With reference now to the figures and in particular to FIG. 2, a block diagram of a data processing system is depicted in accordance with a preferred embodiment of the present invention.Data processing system 200 includes ahost 202, which may include one or more processors, which form the CPU or CPUs fordata processing system 200.Data processing system 200 is a data processing system designed along the Intelligent Input/Output (I₂ O) Architecture Specification, version 1.5, March 1997 available from the I₂ O Special Interest Group, which is incorporated herein by reference. The present invention, however, may be implemented using other system architectures.

The processors withinhost 202 may be, for example, a Pentium II processor operating at 300 Mhz, which is available from Intel Corporation and Santa Clara, Calif. In the depicted example,primary bus 204 andsecondary bus 206 are PCI buses although the present invention may be implemented using other types of buses.

Still referring to FIG. 2,data processing system 200 includes a primary input/output platform (IOP) 208, which is connected to host 202 byprimary bus 204. Additionally,IOP 208 is connected tosecondary bus 206 and also functions as a PCI-to-PCI bus bridge.Data processing system 200 also includesadapter 212 andadapter 214.

Secondary IOPs

210 and 216 are intelligent adapters under I₂ O andsecondary IOP 210 andsecondary IOP 216 contain input/output processors.

Adapters

212 and 214 are non-intelligent adapters, which do not contain input/output processors.

The present invention provides an improved method and apparatus for handling transactions and functions withindata processing system 200 through the use of primary and secondary IOPs. More specifically, communications are provided between primary input/output processor inIOP 208 and the secondary input/output processors in

IOPs

210 and 216. Additionally, the present invention also provides for sharing of tasks or functions betweenprimary IOP 208 and thesecondary IOP 210 orIOP 216.

Turning now to FIG. 3, a block diagram of software units within portions of the data processing system in FIG. 2 is depicted in accordance with a preferred embodiment of the present invention. In the depicted example, the present invention provides communications between devices withindata processing system 200 using a standard architecture for intelligent input/output called the I₂ O Specification by the Intelligent I/O Special Interest Group (I₂ O SIG). This specification defines an architecture for intelligent input/output that is independent of both the specific device being controlled and the host operating system. I₂ O defines a standard architecture for intelligent input/output, an approach to input/output in which low level interrupts are offloaded from the CPU to I/O processors designed specifically to handle input/output. With support for message-passing between multiple independent processors, the I₂ O architecture relieves the host of interrupt-intensive input/output tasks, greatly improving input/output performance in high-bandwidth applications such as networked video, groupware, and client/server processing. I₂ O imposes no restrictions on where layered modules execute, providing support for single processor, multiprocessor and clustered systems. The I₂ O Specification is available from I₂ O Special Interest Group located at 404 Balboa Street, San Francisco, Calif. 94118.Host 202 includes a number of operating system modules (OSMs) 300, 302, 304, 306, 308, and 310. An OS-specific module (OSM) is an upper module that provides the interface to the operating system. Typically, the OS vendor supplies this module, which contains no hardware-specific code. Host 202 also includes amessage layer 312, atransport layer 314, and host operating system 316.Primary IOP 208 includes atransport layer 318, amessage layer 320, intermediate service module (ISM) 322, hardware device module (HDM) #2 324, hardware device module (HDM) #3 326, and bridgeHDM 328. The hardware device module (HDM) is a lower module that provides the interface to the I/O adapter and its devices. The hardware vendor supplies this module, which contains no operating system (OS) specific code. The intermediate service module (ISM) results from splitting the device driver more than once, or adding functionality between the OSM and HDM, creating stackable drivers. This places one or more of these intermediate modules between the OSM and HDM. An independent software vendor can supply ISMs. HDMs and ISMs are often referred to collectively as device driver modules or DDMs, because, in many aspects, their behavior is identical. This is especially true from the viewpoint of the host OS.Bridge HDM 328 also has amessage layer 330 and atransport layer 332 associated therewith.

IOP 210 contains atransport layer 334, amessage layer 336, and aHDM #1 338.IOP 216 contains atransport layer 340, amessage layer 342, and aHDM #4 344. The OSMs withinhost 202 provide an interface to host operating system 316 onhost 202 andmessage layer 312. These OSMs translate requests from host operating system 316 into messages that can be dispatched to the appropriate device driver module (DDM) for processing. Device driver module (DDM) information is forwarded back to host operating system 316 through the OSMs via themessage layer 312.

The HDMs inprimary IOP 208 and secondary IOPs 210 and 216 are the lowest level modules in the depicted example. These HDMs provide the device-specific portion of the device driver that will interface with the particular controller and devices. HDMs are roughly analogous to the hardware-specific portion of the network and SCSI drivers that exist today. The HDM translation layer is unique to each individual hardware device and vendor, and supports a range of operation types, including synchronous and asynchronous requests, and event-driven and polled transactions.

The message layers provide a mechanism for communication between various modules withindata processing system 200 in FIG. 3. These message layers manage and dispatch all requests and provide a set of application programming interfaces (APIs) for delivering messages, along with a set of support routines that process them. Each message layer includes a message handle, a message service routine (MSR), and a message queue. The message queue provides a link between the requestor and the desired service.

The present invention providesbridge HDM 328 includingmessage layer 330 andtransport layer 332 to provide a mechanism for communication with

IOPs

210 and 216. In particular,bridge HDM 328 provides communications betweenprimary IOP 208 and anHDM 338 andHDM 344 in

IOPs

210 and 216, respectively. This bridge HDM executes along with

HDMs

324 and 326 withinprimary IOP 208. In summary,bridge HDM 328 is responsible for managing messaging and transport layers, which interface with all IOPs found onsecondary bus 206. Further, the present invention employs an ISM such asISM 322 to perform functions normally executed onIOP 210 orIOP 216. In this manner, the workload is split between two IOPs, increasing efficiency and performance withdata processing system 200.

Turning now to FIGS. 4A and 4B, are flowcharts of a process for loading and initializing a bridge hardware device module is depicted in accordance with a preferred embodiment of the present invention. The process begins with the intelligent input/output real time operating system (IRTOS) in the primary IOP scanning the secondary bus looking for adapters (step 400). IRTOS is a special purpose real time operating system designed to support high speed, low overhead input/output operations. The IRTOS in the primary IOP looks for IOP devices or adapters on the secondary bus. The process then determines which HDMs to load and initialize by matching an adapter's description to values stored in the HDMs module descriptor tables, which contains information describing the HDM and lists the types of adapters and devices that the HDM can control (step 402).

The bridge HDM initializes and calls i2oDdmCreate() to create a logical configuration table (LCT) entry for itself (step 404). Each entry in an IOP's logical configuration table contains a SubClassInfo word. The structure of this word is defined by each class and identifies the major capabilities of the device. When the DDM registers a device, it provides the information for the LCT entry such as the device's ClassID and SubClassInfo. The IOP publishes this information in its logical configuration table, and the OSM uses this information when it determines which devices to query. Thereafter, the IRTOS sends DdmAdapterAttach requests to the HDM for each IOP found on the secondary PCI bus (step 406). Instep 406, the IRTOS sends a DdmAdapterAttach message to the bridge HDM for each IOP on the secondary bus.

Thereafter, the bridge HDM, in response to receiving the DdmAdapterAttach message initializes the secondary IOP using the I₂ O executive initialization process (step 408). After the secondary IOP has been initialized, the bridge HDM sends an ExecLctNotify request message to the secondary IOP to obtain the LCT for the secondary IOP (step 410). This LCT for the secondary IOP is also referred to as the "secondary LCT". For each device in the secondary LCT that needs to be "bridged", i2oDevCreate() is called to register the device and create a new LCT entry in the primary LCT located in the primary IOP and to set up event handlers for the request messages (step 412). Most of the fields in the new LCT entry would be the same as if they were located in the secondary LCT. For example, fields such as, ClassID and SubClassInfo would be the same.

Next, the IRTOS in the primary IOP determines whether the device identified in the new LCT entry matches the ClassID for which an ISM is registered (step 414). If the device identified in the new LCT entry does not match the ClassID, the process terminates. Otherwise, a determination is made by the IRTOS as to whether the ISM needs to be loaded (step 416). If the ISM is to be loaded, the IRTOS then loads and initializes the ISM (step 418). After the ISM is loaded and initialized, the ISM calls i2oDdmCreate() to the IRTOS (step 420). In response, the IRTOS sends a DdmDeviceAttach message to the ISM for each matching device in the primary LCT (step 422). In response to these requests, the ISM determines whether it is to claim the device identified by the request (step 424). If the ISM claims the device, a UtilClaim is sent by the ISM to the device's target identifier (TID) (step 426).

Thereafter, the bridge HDM translates messages sent to the primary TID and sends them to the secondary TID on the secondary IOP (step 428). The translated message is then passed to the proper TID on the secondary IOP (step 430) with the process terminating thereafter.

With reference now to FIGS. 5A and 5B, are flowcharts of a process for messaging handling is depicted in accordance with a preferred embodiment of the present invention. The operating system module (OSM) creates an I/O request message and passes it to the primary IOP (step 500). The host is the initiator of the message and the device registered by the ISM is the target. This request is sent across the primary PCI bus to the IRTOS in the primary IOP. Thereafter, the IRTOS in the primary IOP routes the request to the proper function handler in the appropriate DDM (step 502). The routing is based on the target address and the function code located in the message header of the request.

The ISM then processes the request (step 504) and determines if a new request should be generated and passed to devices controlled by the bridge HDM (step 506). If a new request is to be generated, the ISM then generates the new request (step 508). The ISM is the initiator of this new request with the HDM device being the target of this new request. This request is then routed by the IRTOS to the proper function handler in the bridge HDM (step 510).

Next, the bridge HDM takes the request and converts it into one that is suitable to be sent to the secondary IOP (step 512). Instep 512, the function handler in the bridge HDM converts the message header to be suitable for passing to a secondary IOP. This conversion is made based on the device context that the IRTOS has passed with the message. At the very least, the initiator address is changed to an I2O HOST TID because the bridge HDM is acting as the host for the secondary IOP. Thereafter, the bridge HDM passes the request to the secondary IOP by reading the inbound FIFO register on the secondary IOP to obtain a request message frame address (MFA) and then copying the request into the message frame and writes the MFA back to the inbound FIFO register (step 514). In assuming the role of the host, the bridge HDM inserts the I₂ O host TID into the initiator field of this request with the target being the HDM in the device (secondary IOP) on the secondary PCI bus.

The IRTOS on the secondary IOP routes the request message to the proper function handler in the HDM on the secondary IOP (step 516). The HDM on the secondary IOP processes the message and takes any action to satisfy the request (step 518). The HDM on the secondary IOP generates an appropriate reply message and passes it to the IRTOS in the secondary IOP (step 520). The host is the initiator of this reply with the target being the HDM in the secondary IOP. In response to receiving the reply, the IRTOS in the secondary IOP routes the reply to its destination (step 522). This destination is the initiator and since the message has an initiator address of I₂ O HOST TID, the IRTOS will route the reply back to its host, which is the primary IOP. This routing is accomplished by copying the reply to a valid reply MFA (memory on the primary IOP) and then writing this MFA to the outbound FIFO register of the primary IOP, which generates an interrupt to the primary IOP. The routing of this message is across the secondary PCI bus to the bridge HDM in the primary IOP.

Thereafter, the bridge HDM services the reply interrupt to obtain the reply, which occurs by reading the outbound FIFO register on the secondary IOP to get the reply MFA (step 524). This reply is converted by the bridge HDM into one suitable for use by the primary IOP (step 526). The ISM is the initiator of this converted reply with the target being the HDM device. The reply is for the request received by the IRTOS back instep 510. The IRTOS then routes the reply to its destination (step 528). In the depicted example, the destination is the ISM in the primary IOP. With reference again to step 506, if the ISM determines that a new request should not be generated, then a reply is built (step 530) with the IRTOS routing the reply back to the host as described instep 528. The reply instep 530 has the host as the initiator and the target as the ISM in the device.

After the IRTOS routes the reply to the ISM, the ISM processes the reply (step 532) and eventually creates a new reply to the original request from the OSM (step 534). With respect to this reply, the initiator is the host with the target being the ISM device. The reply is then routed by the IRTOS to its destination, the host (step 536). This reply is routed across the primary PCI bus. The OSM and the host receives the reply completing the input/output message handling (step 538) with the process terminating thereafter.

Turning now to FIG. 6, an upper level block diagram of a RAID implementation in a data processing system is depicted in accordance with a preferred embodiment of the present invention. Indata processing system 600, tasks are split up between a primary IOP and secondary IOPs to increase performance of the data processing system. In particular, some tasks in the RAID implementation are implemented on the primary IOP rather than solely on a secondary IOP.

In the depicted example,data processing system 600 includes ahost processor 602 in the form of a Pentium II processor connected to aprimary PCI bus 604.Primary IOP 606 also acts as a bus bridge connectingprimary PCI bus 604 tosecondary PCI bus 608. The primary IOP is based on a i960 chip available from Intel Corporation.XOR engine 610 andmemory 612 are connected tosecondary PCI bus 608. Additionally,IOP 614 andIOP 616 also are connected tosecondary PCI bus 608 in the depicted example.IOP 614 andIOP 616 each incorporate a SYMFC920 processor used for PCI to fibre channel I/O. This processor is available from Symbios, Inc., which is located in Fort Collins, Colo. These devices in the depicted example provide a PCI RAID functionality using a fibre channel solution.

In the depicted example,primary IOP 606 includes a RAID ISM to communicate withhost processor 602 while the bridge HDM is used by the RAID ISM to communicate withIOP 614 andIOP 616. As can be seen, in the depicted example, the RAID ISM is placed withinprimary IOP 606 to provide RAID functionality withinprimary IOP 606 rather than leaving it withinIOP 614 orIOP 616. In the depicted example, the fibre channel maintenance is left onIOP 614 andIOP 616. In addition,XOR engine 610 andmemory 612 are used by the RAID ISM located onprimary IOP 606. This architecture allows for transactions betweenhost processor 602 andIOP 614 andIOP 616. In this manner, workload is split between two IOPs rather than being processed by a single IOP, resulting in improved performance. Additionally, bandwidth inprimary PCI bus 604 is increased through the shifting of workload down tosecondary PCI bus 608.

With reference now to FIG. 7, a block diagram of a data processing system is depicted in accordance with a preferred embodiment of the present invention.Data processing system 700 in the depicted example includes two

host processors

702 and 704 running various OSMs that communicate throughprimary PCI bus 706 to one or more of

intelligent adapters

708 and 710 attached toprimary PCI bus 706.

Intelligent adapters

708 and 710 are IOPs incorporating SYMFC920 processors in the depicted example. Additionally, the OSMs may communicate with other intelligent adapters connected tosecondary PCI bus 712 through processor 714, which acts as a bus bridge in addition to providing communications and sharing tasks with secondary input/output processors on

intelligent adapters

716 and 718.

Intelligent adapters

708, 710, 716, and 718 are dedicated to their own hardware devices while processor 714 is a generic processor that can control a number of non-intelligent adapters, such as

adapters

720 and 722 onsecondary PCI bus 712.

The use of a bridge HDM on processor 714 provides for a communications interface betweenprimary PCI bus 706 andsecondary PCI bus 712, allowingsecondary PCI bus 712 to be populated with intelligent adapters along with non-intelligent adapters. Additionally, anadditional processor 724 may be employed to provide communication betweensecondary PCI bus 712 andsecondary PCI bus 726 in whichnon-intelligent adapter 728 and anintelligent adapter 730 are located. As with processor 714,processor 724 also includes a bridge HDM to enable the communications between the two secondary PCI buses. In this manner, ISMs may be implemented inprocessors 714 and 724, which functions with the intelligent adapters on the secondary PCI buses.

Turning to FIG. 8, a block diagram of a bus bridge is depicted in which a preferred embodiment of the present invention may be implemented. IOP/bus bridge 800 is an example of an IOP/bus bridge, such as IOP/bus bridge 714 or 724 in FIG. 7. IOP/bus bridge 800 includes aprocessor core 802 in communications with alocal bus 804, which connectsprocessor core 802 to a number of components within IOP/bus bridge 800, such as I/O APIC interface 806,PC interface 808,memory controller 810, internallocal bus arbiter 812, and secondaryPCI arbitration unit 814.Processor core 802 also is coupled to interruptrouting unit 812. IOP/bus bridge 800 also includes a twochannel DMA controller 818, a primaryaddress translation unit 820, and amessaging unit 822 all having connections tolocal bus 804 and primaryinternal PCI bus 824. A secondaryaddress translation unit 826 and a onechannel DMA controller 828 both have a connection tolocal bus 804 and secondaryinternal PCI bus 830. Primaryinternal PCI bus 824 and secondaryinternal PCI bus 830 are in communication withprimary PCI bus 832 andsecondary PCI bus 834, respectively. PCI toPCI bridge unit 836 provides communication betweenprimary PCI bus 832 andsecondary PCI bus 834. A primaryinternal arbiter PCI 838 and a secondaryinternal PCI arbiter 840 also are located within IOP/bus bridge 800. In the depicted example, IOP/bus bridge 800 is representative of an i960 RD I/O processor chip available from Intel Corporation.

Thus, the present invention provides an improved method and apparatus for processing input/output transactions within a data processing system. This advantage is provided by placing intelligent adapters, which may be in the form of IOPs containing input/output processors, on a secondary bus that communicates with a primary IOP that is connected to a primary bus and the secondary bus. The primary IOP also acts as a bus bridge for a primary bus and a secondary bus. Functionality normally performed solely by IOPs on the secondary bus may be placed also within the primary IOP to split up workloads and increase performance on the data processing system.

Additionally, the present invention provides a shifting of workload to the secondary bus, which increases the bandwidth within the data processing system. Furthermore, communications between secondary IOPs and the primary IOP are set up such that the secondary IOPs see the primary IOP as the host processor. This feature is provided in the depicted example by placing a host identifier in messages generated by the primary IOP and sent to the secondary IOP. In turn, these messages are identified such that replys to these messages from the secondary IOP, which are directed to the host, are routed to the primary IOP for processing.

The description of the preferred embodiment of the present invention has been presented for purposes of illustration and description, but is not limited to be exhaustive or limited to the invention in the form disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art. For example, although the terminology was described in terms of the Intelligent Input/Output Architecture Specification and involved PCI buses, the present invention may be applied to other bus architectures involving primary and secondary buses. More specifically, the present invention may be applied to more intelligent adapters other than IOPs. The embodiment was chosen and described in order to best explain the principles of the invention the practical application to enable others of ordinary skill in the art to understand the invention for various embodiments with various modifications as are suited to the particular use contemplated.

Claims

What is claimed is:

1. A data processing system comprising:

a primary bus;

a secondary bus;

a host processor connected to the primary bus;

a first secondary processor connected to the primary bus and the secondary bus, wherein the first secondary bus processor provides communication between the primary bus and the secondary bus; and

a second secondary processor connected to the secondary bus, wherein a plurality of functions are performed by the second secondary processor and wherein a portion of the functions within the plurality of functions are assigned to the first secondary processor for performance by the first secondary processor such that performance of the portion of the functions by the second secondary processor is absent.

2. The data processing system of claim 1, wherein the primary bus is a PCI bus and the secondary bus is a PCI bus.

3. The data processing system of claim 1, wherein the first secondary processor is within an input/output platform.

4. The data processing system of claim 3, wherein the second secondary processor is within an input/output platform.

5. The data processing system of claim 1, wherein the data processing system is an intelligent input/output data processing system under an intelligent input/output architecture specification.

6. The data processing system of claim 1, wherein the first secondary processor is within an input/output platform.

7. The data processing system of claim 6, wherein the second secondary processor is within an input/output platform.

8. A data processing system comprising:

a primary bus;

a secondary bus;

a host processor connected to the primary bus;

a first secondary processor connected to the primary bus and the secondary bus, wherein the first secondary processor provides communication between the primary bus and the secondary bus;

a second secondary processor connected to the secondary bus; and

wherein a first device driver and a second device driver define functions performed by the second secondary processor within an input/output platform and wherein the first device driver is loaded on an input/output platform containing the first secondary processor and the second device driver is loaded on the input/output platform containing the second secondary processor such that the first secondary processor performs a function defined by the first device driver and the second secondary processor performs a function defined by the second device driver.

9. The data processing system of claim 8, wherein the first device driver is an intermediate service module.

10. The data processing system of claim 9, wherein the second device driver is a hardware device module.

11. A data processing system comprising:

a primary bus;

a host processor connected to the primary bus;

a secondary bus;

a bus bridge connecting the primary bus to the secondary bus, wherein the bus bridge includes:

a processor;

an input/output platform connected to the secondary bus;

first message means for sending messages to the input/output platform from the bus bridge, wherein a message generated by the processor in the bus bridge includes an identifier that identifies the message as originating from the host processor; and

second message means for identifying messages to the host processor that originated from the bus bridge and routing the messages to the bus bridge.

12. The data processing system of claim 11 further comprising a second input/output platform connected to the primary bus.

13. A data processing system comprising:

a primary bus;

a host processor connected to the primary bus;

a secondary bus;

a processor;

an input/output platform connected to the secondary bus;

first message means for sending messages to the input/output platform from the bus bridge, wherein a message generated by the processor in the bus bridge includes an identifier that identifies the message as originating from the host processor;

second message means for identifying messages to the host processor that originated from the bus bridge and routing the messages to the bus bridge; and

means for shifting workload from the primary bus to the secondary bus, wherein bandwidth is increased.

14. A data processing system comprising:

a primary bus;

a host processor connected to the primary bus;

a secondary bus;

a processor;

an input/output platform connected to the secondary bus;

second message means for identifying messages to the host processor that originated from the bus bridge and routing the messages to the bus bridge;

a plurality of functions are performed by the input/output platform; and

assignment means for assigning a portion of the plurality of functions to the processor in the bus bridge, wherein workload is divided between the bus bridge and the input/output platform.

15. The data processing system of claim 14, wherein the portion of the plurality of functions are embodied in a device driver and wherein the assignment means loads the device driver on to the bus bridge.

16. The data processing system of claim 15, wherein the device driver is an intermediate service module.

17. The data processing system of claim 14, wherein the portion of the plurality of functions is a RAID function.

18. A data processing system comprising:

a primary bus;

a host processor connected to the primary bus;

a secondary bus;

a processor;

an input/output platform connected to the secondary bus;

a second secondary bus; and

a second bus bridge, wherein the second bus bridge provides communications between the first secondary bus and the second secondary bus.

19. The data processing system of claim 18 further comprising:

a second input/output platform connected to the second secondary bus,

third message means for sending messages to the input/output platform from the bus bridge and in which a message generated by the processor in the bus bridge includes an identifier that identifies the message as originating from the host processor; and

fourth message means for identifying messages to the host processor that originated from the bus bridge and routing the messages to the bus bridge.

20. A bus bridge for providing communications between a primary bus and a secondary bus, the apparatus bus bridge comprising:

a processor including first instruction means for sending a message to an input/output platform connected to the secondary bus, wherein the message identifies a host processor as an originator of the message; second instruction means for identifying message targeting the host processor as a reply to a message originated by the processor; and third instruction means, for routing the message, identified by the second instruction means, to the processor.

21. A bus bridge for providing communications between a primary bus and a secondary bus, the apparatus bus bridge comprising:

a processor including first instruction means for sending a message to an input/output platform connected to the secondary bus, wherein the message identifies a host processor as an originator of the message;

second instruction means for identifying message targeting the host processor as a reply to a message originated by the processor;

third instruction means, for routing the message, identified by the second instruction means, to the processor; and

fourth instruction means for performing a function for the input/output platform, wherein instruction means for performing the function on the input/output platform are absent.

22. The bus bridge of claim 21, wherein the fourth instruction means is a device driver.

23. The bus bridge of claim 22, wherein the device driver is an intermediate service module.